In Orthogonal Frequency Division Multiplexing(OFDM) system, super resolution time delay estimation shows important application value. The Multiple Signal Classification(MUSIC) algorithm requires eigenvalue decomposition whose computational complexity is very high. A new time delay estimation based on Propagator Method(PM) is proposed for solving the problem. Firstly, the proposed algorithm performs the channel frequency response estimation in OFDM system. Moreover, the channel frequency response estimation can be used for estimating the covariance matrix which can calculate the propagator. The propagator can be used to construct the noise subspace without eigenvalue decomposition, then the noise subspace is orthonormalized for improving the performance. Finally, the pseudo spectral function is constructed for estimating the time delay. Simulation results and the analysis of computational complexity show that the proposed algorithm has the similar performance as the MUSIC algorithm and approaches the Cramer-Rao Bound(CRB) under the condition of the significant decline of the computational complexity.